It's been some time since I last discussed Conformity Assessment and with good reason. With the advent of the update to the EASA Basic Regulation, the process is expected to change. This change may well overhaul the Conformity Assessment process entirely and address many of the challenges faced today. But this blog is not intended to discuss this, rather it postulates a long-term vision for the Conformity Assessment process in the future. What could it look like in 20, 30 or even 40 years?
When I thought about this question I could not let go of the idea that the process, in its current state, relies on interactions between numerous stakeholders even for the simplest of system changes. Of course, at a high level there is the National Supervisory Authority (NSA), Air Navigation Service Provider (ANSP) and the manufacturer, but there are also internal stakeholders. The ANSP's role alone involves Project Managers and several Subject Matter Experts (SMEs) responsible for different aspects of the implementation. Completing the Conformity Assessment process can therefore be challenging for staff who are constrained by the demands of several concurrent projects, limited resources and a lack of familiarity with what the process involves. It is only natural to wonder how, or even if, these challenges can be overcome.
Well perhaps the answer is to turn to technology. Could such a process ever be automated? The default response is to say it is too difficult, too complex a problem to entrust to computers over humans, and takes too much time to implement. However, with the emergence of simple but powerful data analysis techniques such as natural language processing and machine learning, these concerns could arguably soon be allayed. Since Conformity Assessment is the process of gathering and documenting evidence to demonstrate compliance with the requirements of the Interoperability Regulation and associated Implementing Rules (IRs), it seems an ideal candidate for the virtues of such techniques. Computer programs can learn to recognise the documents in which applicable evidence could be found, and to pair such evidence with appropriate requirements. They could even learn to recognise applicable means of compliance and evaluate a project's compliance with these. To resort to hyperbole, the possibilities for each stakeholder are endless.
I am aware that my suggestion glosses over some of the key challenges faced today by those involved in Conformity Assessment that may remain in the future. For example, current requirements are highly open to interpretation and difficult to apply, and there are not always clear Means of Compliance and Community Specifications. But maybe there could even be a role for computers here. Could the techniques we've discussed help to achieve pan-European harmonisation of the process through understanding each Member States' interpretations of Interoperability requirements?
So, what is the long-term future of Conformity Assessment? I firmly believe that despite some initial hurdles, automation is the answer.
1: The ability of a computer to recognise the meaning, context and connotations of language.
2:The ability of a computer to "learn through experience" – to recognise significant features of a dataset based on previous information processed by the program.